[Insight-users] Using image iterators combined with filters in a loop

[Merrem, Andreas]

 Hi Luis, 

The DisconnectPipeline method worked, even in combination with the iterator. 
Thanks a lot for taking the time to help, it was really essential for the project I'm working on. 

Regards, 

Andreas


-----Ursprüngliche Nachricht-----
Von: Luis Ibanez [mailto:luis.ibanez at kitware.com] 
Gesendet: Mittwoch, 26. Mai 2010 21:49
An: Merrem, Andreas
Cc: insight-users at itk.org
Betreff: Re: Using image iterators combined with filters in a loop

Hi Andreas,

Thanks for your detailed description of the problem.

If you want to implement an iteration loop over a sub-pipeline you should disconnect the output of the filter.

For example a pipeline of filter A, B, C, run in a loop should look like:


   ImageType::Pointer image = GetInputImage();

   for( int i = 0; i < N;  i++ )
     {
     A->SetInput( image );
     B->SetInput( A->GetOutput() );
     C->SetInput( B->GetOutput() );
     C->Update();
     image = C->GetOutput();
     image->DisconnectPipeline();
     }


If you don't call DisconnectPipeline() you will create a circular dependency between inputs and outputs in the A,B,C pipeline.


Also, you could replace the manual use of ImageIterators with the filter:

Insight/Code/Review/
     itkMultiplyByConstantImageFilter.h


http://www.itk.org/Doxygen/html/itkMultiplyByConstantImageFilter_8h.html




   Regards,


        Luis



---------------------------------------------------------
On Wed, May 26, 2010 at 2:17 PM, Merrem, Andreas <andreas.merrem at medma.uni-heidelberg.de> wrote:
>
>
> -----Ursprüngliche Nachricht-----
> Von: Merrem, Andreas
> Gesendet: Mittwoch, 26. Mai 2010 20:08
> An: 'Luis Ibanez'
> Cc: insight-users at itk.org
> Betreff: Using image iterators combined with filters in a loop
>
>
> Hello Luis (or anyone else who can help),
>
> Thanks a lot for your last answer to my registration question.
> Right now I'm stuck with another problem. I am writing a registration program that involves using image iterators and filters within a loop. To test some parts of what I want to do, I have written a test program that should read two images and process them with a loop. I want to multiply all pixel values by a factor and then smooth one of them during each loop iteration.
>
> This way, the program compiles, but it doesn't work the way it should (the first output image is the input image multiplied by the factor only once and then smoothed).
> It does work the way it should when I don't use a loop, copy and paste the content of the loop e.g. 3 times, and instantiate a both a new iterator and a new smoothing filter for each "iteration".
> However, I can't do that for an arbitrary number of iterations (I would have multiple definitions).
>
> Is there a way to use one smoothing filter and one iterator for my program?
> My short program is below this message, please ignore the shuffling around with the vector image, it was just a test, and it has nothing to do with the problem.
>
> Thank you very much for your help,
>
> Andreas
>
>
> //Test Program: smoothing vector images
>
> #include "itkImage.h"
> #include "itkImageFileReader.h"
> #include "itkImageFileWriter.h"
> #include "itkSmoothingRecursiveGaussianImageFilter.h"
> #include "itkCovariantVector.h"
> #include <iostream>
>
>
> int main( int argc, char * argv[] )
> {
>
>
> //Read Images
>
> const    unsigned int    Dimension = 2; typedef  float   PixelType;
>
> typedef itk::Image <PixelType, Dimension> FixedImageType; typedef 
> itk::ImageFileReader <FixedImageType> FixedImageReaderType;
>
> FixedImageReaderType::Pointer fixedImageReader1  = 
> FixedImageReaderType::New(); FixedImageReaderType::Pointer 
> fixedImageReader2  = FixedImageReaderType::New();
>
>
> fixedImageReader1-> SetFileName (argv[1]);
> fixedImageReader2-> SetFileName (argv[2]);
>
>  try
>    {
>    fixedImageReader1-> Update();
>        fixedImageReader2-> Update();
>
>    }
>  catch (itk::ExceptionObject & e)
>    {
>    std::cerr << "exception in file reader " << std::endl;
>    std::cerr << e << std::endl;
>    return EXIT_FAILURE;
>    }
>  FixedImageType::Pointer FixedImage1 = fixedImageReader1-> 
> GetOutput();
>  FixedImageType::Pointer FixedImage2 = fixedImageReader2-> 
> GetOutput();
>
> //Create Vector Image
>
>  typedef float VectorComponentType;
>  typedef itk::CovariantVector <VectorComponentType, 2> VectorType;
>  typedef itk::Image <VectorType, Dimension> VectorImageType;
>  VectorImageType::Pointer VectorImage = VectorImageType::New();
>  VectorImage-> SetRegions (FixedImage1-> GetLargestPossibleRegion());
>  VectorImage-> Allocate();
>
>
>
> typedef itk::ImageRegionIterator <VectorImageType> VecIteratorType; 
> VecIteratorType vecIt (VectorImage, VectorImage-> 
> GetLargestPossibleRegion());
>
> typedef itk::ImageRegionIterator <FixedImageType> FixIteratorType; 
> FixIteratorType fixIt1 (FixedImage1, FixedImage1-> 
> GetLargestPossibleRegion()); FixIteratorType fixIt2 (FixedImage2, 
> FixedImage2-> GetLargestPossibleRegion());
>
>
>
> VectorType pix;
>
> float L=1.3;
>
>
>  //Smoothing Filter (which doesn't work)
>
>  typedef itk::SmoothingRecursiveGaussianImageFilter <FixedImageType, 
> FixedImageType> SmoothFilterType;
>  SmoothFilterType::Pointer Smoother1 = SmoothFilterType::New();
>  Smoother1-> SetNormalizeAcrossScale (false);
>  Smoother1-> SetSigma (1.0);
>
>
>  //Output Image from Smoother, and iterator to access it (the one that 
> doesn't work)
>
>  FixedImageType::Pointer SmoothIm = FixedImageType::New();
>  SmoothIm=FixedImage1;
>  FixIteratorType smoothIt1 (SmoothIm, 
> SmoothIm->GetLargestPossibleRegion());
> // the largest possible region for the smoother output image is the 
> same as for the fixed (input) image
>
>  /////////////////////////////Loop for 
> processing/////////////////////////////
>
> int n=3;
>
> for (int i=0; i<n; i++){
>
>  //Fill Vector Image
>  for ( fixIt1.GoToBegin(), fixIt2.GoToBegin(), vecIt.GoToBegin(); 
> !vecIt.IsAtEnd(); ++fixIt1, ++fixIt2, ++vecIt)
>    {
>                pix[0] = fixIt1.Get();
>                pix[1] = fixIt2.Get();
>                vecIt.Set (pix);
>    }
>  //Vector Image*L
>  for (vecIt.GoToBegin(); !vecIt.IsAtEnd(); ++vecIt)
>    {
> vecIt.Set (L*vecIt.Get());
>    }
> // Back to component images
>
>  for ( fixIt1.GoToBegin(), fixIt2.GoToBegin(), vecIt.GoToBegin(); 
> !fixIt1.IsAtEnd(); ++fixIt1, ++fixIt2, ++vecIt)
>    {
>                fixIt1.Set(vecIt.Get()[0]);
>                fixIt2.Set(vecIt.Get()[1]);
>    }
>
>
>  //Copy and smooth first component (this is where the problem is)
>
>
>
>  Smoother1-> SetInput (FixedImage1);
>  Smoother1-> Update();
>  SmoothIm = Smoother1-> GetOutput();
>
>  for ( fixIt1.GoToBegin(), smoothIt1.GoToBegin(); !fixIt1.IsAtEnd(); 
> ++fixIt1, ++smoothIt1)
>  {
>          fixIt1.Set(smoothIt1.Get());
>  }
>
>
> }
> //////////////////////////////////////////////////////////////////////
>
>   //Write Images to File
>
>
>
>  typedef unsigned char OutputPixelType;
>  typedef itk::Image <OutputPixelType,Dimension> OutputImageType;
>
>  typedef itk::ImageFileWriter <OutputImageType> WriterType;
>  WriterType::Pointer Writer1 = WriterType::New();
>  WriterType::Pointer Writer2 = WriterType::New();
>
>  typedef itk::CastImageFilter <FixedImageType,OutputImageType> 
> CastFilterType;
>  CastFilterType::Pointer  caster1 =  CastFilterType::New();
>  CastFilterType::Pointer  caster2 =  CastFilterType::New();
>
>  caster1-> SetInput (FixedImage1);
>  caster2-> SetInput (FixedImage2);
>  Writer1-> SetInput (caster1-> GetOutput());
>  Writer2-> SetInput (caster2-> GetOutput());
>  Writer1-> SetFileName (argv[3]);
>  Writer2-> SetFileName (argv[4]);
>
>  Writer1-> Update();
>  Writer2-> Update();
>
>  return EXIT_SUCCESS;
>  }
>
>
>
>